Sustaining Learning Investment amid Schooling Disruptions? Evidence from Online Tutoring in the West Bank

Voluntary remedial education programs face two compounding challenges: attracting and retaining students who are not required to attend, and ensuring that participation — when it happens — actually improves learning. In fragile and conflict-affected settings, both challenges are more acute: recurrent disruptions erode the routines and social ties that sustain engagement, while the cumulative learning losses they produce make remediation more urgent.

In the West Bank, for example, approximately 602,000 public school students learn in a hybrid system combining in-person and online instruction. The Ministry of Education and Higher Education (MOEHE) has developed digital infrastructure to mitigate the resulting loss of instructional time, but platform utilization remains low and learning gaps persist. By 2024, only 7 percent of children in need of remedial education and catch-up classes had been reached. These disparities underscore the need for virtual remedial education targeting the most vulnerable students while guaranteeing that students remain engaged into the virtual remedial programs.

This projects aims to evaluate an online tutoring program delivered by the MOEHE for lagging Grade 7-9 students in the West Bank, where in-person schooling has been reduced to two days per week since October 2023, and use it as a platform to test whether (and for how long) different sources of motivation sustain voluntary engagement in learning.

Four thousand students in vulnerable schools will be randomly assigned at the individual level to one of four arms: a control group; standard small-group online tutoring (36 hours over 12 weeks, delivered via Microsoft Teams); tutoring plus an internal regulation component (a five-minute group cohesion activity at the start of each session, with no academic content); or tutoring plus an external regulation component (a five-minute collaborative problem set at the start of each session, with group-contingent rewards). Instructional time, dose, subject coverage, and delivery mode are held constant across the three tutoring arms and only the motivational scaffold varies.

Primary outcomes to be measured are take-up, within-session engagement, and foundational learning in mathematics and Arabic language. Secondary outcomes include students' social cohesion, school attendance during the program, mental health, and persistence of academic performance and engagement after program ends. Primary and secondary outcomes, as well as other variables for heterogeneity analysis, will be collected through three survey rounds with students, teachers and tutors (baseline, midline, and endline). Outcomes related to within-session engagement will be measured using session-level monitoring data reported by tutors.

The design has three features that distinguish it from existing tutoring evaluations. First, engagement is measured at the session level for all 36 sessions rather than proxied by attendance, allowing exposure to be decomposed into showing up and participating. Second, a follow-up survey four months after rewards are withdrawn tests the crowding-out prediction, on which self-determination theory and standard incentive models make opposite-signed forecasts. Third, session-level engagement data crossed with locality-time variation in school closures and mobility restrictions permits an event-study analysis of whether group cohesion buffers learning engagement against conflict-related shocks.